Korea AI Basic Act and Enforcement Decree

MSIT states that the Basic Act on the Development of Artificial Intelligence and the Establishment of a Trustworthy Foundation comes into effect on 22 January 2026, together with its Enforcement Decree after Cabinet approval on 20 January 2026.

MSIT also states that a grace period of at least one year will defer fact-finding investigations and penalties except in exceptional cases involving serious social harm. Legal force and immediate penalty exposure are therefore not the same thing in early 2026.

According to MSIT, Enforcement Decree Article 23 requires operators using high-impact or generative AI to inform users in advance that AI is being used.

For generative AI outputs, including deepfakes, the decree requires clear labelling recognisable to users, taking into account factors such as age. For some non-deepfake AI-generated content, MSIT says invisible digital identifiers such as watermarks may be used.

MSIT describes a cumulative training-compute threshold above 10^26 FLOPs, together with state-of-the-art technology and broad fundamental-rights impact, as the criteria stack for safety obligations under Enforcement Decree Article 24, with detailed determination guidance to follow in guidelines.

• Advance notice that high-impact or generative AI is in use (Art. 23 as described by MSIT).
• Clear labelling of generative outputs including deepfakes.
• Safety stack around >10^26 FLOPs plus additional criteria (Art. 24 as described by MSIT).
• Grace period ≥1 year for most penalties except serious-harm cases.

The Act combines promotion and trustworthiness duties. It should not be collapsed into a pure labelling statute or a pure compute-licensing regime without the Korean operative text.

Article numbers cited here are MSIT’s English description of the Enforcement Decree. Where English and Korean diverge, the Korean Official Gazette controls.

This briefing follows MSIT’s English announcement. The Korean Official Gazette remains authoritative for litigation. Later MSIT guidelines on safety determination and labelling presentation will refine how Articles 23 and 24 operate in practice.

The following actors are linked to this vehicle in the graph. Their presence here records institutional role on the page, not an independent finding that each actor has completed a compliance duty.

• Ministry of Science and ICT (Korea) (government): Lead Korean ministry implementing the AI Basic Act, Enforcement Decree and related guidance.

Each block below is an explicit evidence row: a claim that points to one supporting passage. Read the quote and locator before treating the paraphrase as settled.

• Korea’s AI Basic Act took effect on 22 January 2026. The announcement fixes the operative date and notes a grace period of at least one year for most penalties. Operational note: It establishes Asia’s first comprehensive AI statute as current law subject to grace-period administration.
• Korea’s Enforcement Decree requires clear labelling of generative AI outputs. The English release attributes the labelling mechanics to the Enforcement Decree accompanying the AI Basic Act. Operational note: It adds a statutory Korean vehicle to the labelling ask alongside EU and Chinese measures.

Dated events attached to this vehicle. Each entry is a graph event with its own update page when present.

• 2026-01-22 · effective-date · Korea AI Basic Act enters into force. The effective date started operative transparency and safety duties while deferring most penalties. Consequence recorded: Brought Asia’s first comprehensive AI statute into force.

A vehicle is an instrument. Asks are the outcomes the instrument is treated as advancing in the AI Tribune graph.

• Require visible disclosure for deepfakes and public-interest synthetic media Status: in-force. Require people encountering covered synthetic media to receive a clear, visible disclosure rather than relying only on hidden metadata.
• Require advance notice and clear labelling for generative and high-impact AI Status: in-force. Require AI business operators providing high-impact or generative AI to inform users in advance and clearly label generative outputs, including deepfakes.

These documents are the originals behind the evidence rows. Prefer the document text over secondary paraphrase when the two diverge.

• The AI Basic Act Comes into Force to Lay the Foundation for Korea to Become an AI G3 (Official press release, 2026-01-22). MSIT English announcement that the AI Basic Act and Enforcement Decree take effect on 22 January 2026, with transparency and safety duties.